Reliability Tests
We present the testing results to show MegaMatcher Accelerator 13.1 fingerprint template matching engine reliability.
These fingerprint image datasets was used for testing:
- NIST Special Database 14 (SD14), which consists of 54,000 fingerprint images, corresponding to 27,000 unique fingers from 2,700 subjects. Information about the SD14 is no longer available, but you may check the archived copy for general information.
- Neurotechnology's 100,000,000 artificial fingerprint images dataset. Download a small sample set.
The testing results are presented as a receiver operation characteristic (ROC) chart. ROC show the dependence of false rejection rate (FRR) on the false acceptance rate (FAR). The ROC calculation was performed by comparing all templates from the SD14 to to a gallery database, composed from a combination of both datasets.
The tests were performed with these biometric template types:
- 1 fingerprint record – the gallery represented 100,027,000 subjects. MegaMatcher Accelerator 13.1 algorithm reliability in this test is shown as red curve on the ROC chart.
- 2 fingerprint records – the gallery represented 50,013,500 subjects. MegaMatcher Accelerator 13.1 algorithm reliability in this test is shown as blue curve on the ROC chart.
- 4 fingerprint records – the gallery represented 25,006,750 subjects. MegaMatcher Accelerator 13.1 algorithm reliability in this test is shown as green curve on the ROC chart.
- 8 fingerprint records – the gallery represented 12,503,375 subjects. This test resulted in 0% false rejection rate for all false acceptance rate settings, and the reliability in this test is shown as magenta dot on the ROC chart.
